2012 © Trivadis BASEL BERN LAUSANNE ZÜRICH DÜSSELDORF FRANKFURT A.M. FREIBURG I.BR. HAMBURG MÜNCHEN STUTTGART WIEN Welcome November 2012 Columnstore Indexes and Partition Improvements 1 SQL Server 2012 Columnstore Indexes and Table Partitioning Improvements November 2012 Meinrad Weiss

2012 © Trivadis Column Store: Data Warehouse Query Performance Columnstore indexes provide an easy way to significantly improve data warehouse and decision support query performance against very large data sets  Ideal candidates include queries against star schemas that use filtering, aggregations and grouping against very large fact tables Performance improvements for “typical” data warehouse queries from 10x to 100x 2 November 2012 Columnstore Indexes and Partition Improvements

2012 © Trivadis DW Queries in SQL Server 2008 and SQL Server 2012 You need to execute high performance DW queries against very large data sets?  In SQL Server 2008 and SQL Server 2008 R2  OLAP (SSAS) MDX solution  ROLAP and T-SQL + intermediate summary tables, indexed views and aggregate tables -(Inherently inflexible)  In SQL Server 2012  You can create a columnstore index on a very large fact table referencing all columns with supporting data types -Utilizing T-SQL and core Database Engine functionality -Minimal query refactoring or intervention 3 November 2012 Columnstore Indexes and Partition Improvements SQL MDX - Dimensions - Key Figures - Aggregates (Speed) - Link to Excel - Data Mining

2012 © Trivadis Column vs. Row Store November 2012 Columnstore Indexes and Partition Improvements 4 Row Store (Heap / B-Tree) ProductIDOrderDateCost ’ ’ ’ data page “1000” data page “1001” ProductIDOrderDateCost ’ ’ ’ ProductID OrderDate Cost 2’ ’ ’ ’ ’ ’ data page “2000” data page “2001” data page “2003” Column Store (values compressed)

2012 © Trivadis Creating a columnstore index 5 T-SQL SSMS November 2012 Columnstore Indexes and Partition Improvements

2012 © Trivadis Good Candidates for Columnstore Indexing Table candidates:  Very large fact tables (for example – billions of rows)  use more or less all attributes -Adding additional columns will only have an effect at creation time  (Larger dimension tables (millions of rows) with compression friendly column data)  Update of rows may get quite complex  If unsure, it is easy to create a columnstore index and test the impact on your query workload 6 November 2012 Columnstore Indexes and Partition Improvements Fact Table Many rows Few attributes Fact Table Column Store Index

2012 © Trivadis Supported Data Types Supported data types  Char / nchar / varchar / nvarchar  (max) types, legacy LOB types and FILESTREAM are NOT supported  Decimal/numeric  Precision greater than 18 digits NOT supported  Tinyint, smallint, int, bigint  Float/real  Bit  Money, smallmoney  Date and time data types  Datetimeoffset with scale > 2 NOT supported 7 November 2012 Columnstore Indexes and Partition Improvements

2012 © Trivadis Maintaining Data in a Columnstore Index Once built, the table becomes “read-only” and INSERT/UPDATE/DELETE/MERGE is no longer allowed ALTER INDEX REBUILD / REORGANIZE not allowed Other options are still supported:  Partition switches (IN and OUT)  Drop columnstore index / make modifications / add columnstore index  UNION ALL (but be sure to validate performance) 8 November 2012 Columnstore Indexes and Partition Improvements

2012 © Trivadis Working with Partitions SQL Server 2012 supports up to 15’000 partitions per table Nice new Wizard November 2012 Columnstore Indexes and Partition Improvements 9 Partitioned Table (read only) Range 2012_08 Range 2012_09 Range 2012_10 Range 2012_11 Table Partitioned Table (read only) Switch In ETL Column Store Index Read-only Table Create Column Store index Create Column Store index

2012 © Trivadis BASEL BERN LAUSANNE ZÜRICH DÜSSELDORF FRANKFURT A.M. FREIBURG I.BR. HAMBURG MÜNCHEN STUTTGART WIEN Let‘s go. November 2012 Columnstore Indexes and Partition Improvements 10 Wettbewerb